I'm not using scsi, I'm using the onboard IDE interface... I've said that in the past several notes. Looking at this again, it appears like it might be load related as well and not a total fix. For one thing, it is _still_ continually going and trying to hit that other disk and saying it has failed and is in degraded mode. This doesn't appear to be hurting anything, but it's not clear why it keeps trying to use the disk once it knows it has failed. I just tried the same test under much heavier loading, and unfortunately it again failed. It's possible it's random luck as well - i.e. how I pull the power, noise, etc. In today's case, the moment I pull the power, all disk I/O ceases, but if I plug the power back in, it instantaneously recovers into degraded mode. Sortof like it needs to talk to the drive after it as failed for just a tiny bit of communication, and once it has done that, it recovers. I'm not really familiar with the raid code in the kernel, but is the md driver clearing the write cache and removing any pending I/O operations against the given block device when a device failure occurs? That might help if it isn't doing that currently. -- Nathan ------------------------------------------------------------ Nathan Neulinger EMail: [EMAIL PROTECTED] University of Missouri - Rolla Phone: (573) 341-4841 Computing Services Fax: (573) 341-4216 > -----Original Message----- > From: Osma Ahvenlampi [mailto:[EMAIL PROTECTED]] > Sent: Thursday, June 17, 1999 1:41 AM > To: Neulinger, Nathan R. > Cc: [EMAIL PROTECTED] > Subject: Re: Success! 2.2.6 appears to fix the lockup problem I was > having wit h RAID1 > > > Neulinger, Nathan R. <[EMAIL PROTECTED]> writes: > > I decided to try out 2.2.6, and it appears to solve the > problem I was having > > with the system never recovering when I pulled the power on > one of the > > drives in the mirror set. I was using 2.2.2. > > Could you repeat what SCSI controller you're using? If 2.2.6 really > did fix it for you (and you weren't just lucky this time), it > would be > interesting to know which driver has been improved to allow this.. > > -- > Osma Ahvenlampi >
